The Knowledge Jar AI Community Rewind: April 14, 2024

Welcome back to The Knowledge Jar AI Community!

In this update, we will recap last week's events and preview this week's events. Additionally, we will give you a brief summary of articles related to AI. By joining our community, you will have access to webinars and tutorials and can stay informed about upcoming in-person events.

Last Weeks Events

Tutorials

This Weeks Events

Are you a small business owner who's heard about AI but isn't sure what it means for your company? You're not alone! There's a lot of hype and fear about artificial intelligence, but Victoria Winter is here to tell you that AI could be the greatest gift to small businesses - if you know how to harness its power. Join the conversation this Wednesday.

AI News

Meta has confirmed that it plans to release Llama 3, the next generation of its large language model, within the next month. The release of Llama 3 will include several different models with varying capabilities and versatility, as Meta aims to catch up to OpenAI's ChatGPT and power multiple products across the company Learn More

Nvidia's dominance in the AI chip market is being challenged by tech giants like Google, Intel, Meta, Microsoft, and Amazon, who are all developing their own AI hardware to power their generative AI models and services. However, Nvidia won't give up its market share easily, as it continues to innovate and maintain its technological edge over the competition. Learn More

Google has announced the Axion, its first custom Arm-based CPU designed for the data center, which will deliver industry-leading performance and energy efficiency. The Axion CPU, based on the Arm Neoverse V2 platform, will be available to Google Cloud customers later this year and offers 30% better performance than the fastest general-purpose Arm-based instances and 50% better performance than comparable x86-based VMs. Learn More

Google has made the second generation of its Imagen AI model, which can create and edit images from text prompts, more widely available to Google Cloud customers using Vertex AI. Imagen 2 has significantly improved image quality compared to the first-generation model, and introduces new capabilities such as the ability to render text and logos in multiple languages. Learn More

Google has launched a new AI-powered video creation app called "Google Vids" that is designed to simplify and streamline the process of creating informative video content for Google Workspace users, leveraging the capabilities of Google's Gemini AI technology. Google Vids allows users to manually storyboard with their own media files or utilize AI to automatically craft storyboards based on keywords and prompts, and includes features like a voice-over function and a library of stock videos, images and background music to facilitate efficient video production, especially for corporate use cases like training, sales, and project updates. Learn More

OpenAI reportedly transcribed over a million hours of YouTube videos to train its advanced GPT-4 language model, which falls into a legally questionable gray area around AI copyright law. Both Google and YouTube have stated that unauthorized scraping or downloading of YouTube content is prohibited by their terms of service, indicating that OpenAI's actions may have violated these policies. Learn More

The U.S. Congress has introduced a new bill called the "Generative AI Copyright Disclosure Act" that would require AI companies like OpenAI and Google to disclose the copyrighted material they use to train their large language models, which power generative AI assistants like ChatGPT and Bard. The legislation aims to bring more transparency around the use of copyrighted data in training these AI models, as there have been concerns about potential copyright infringement. Learn More

Adobe is scrambling to build its own text-to-video AI model to compete with rivals like OpenAI's Sora, and is buying up large amounts of video footage to train this new AI model. To keep its AI tools legally and commercially safe, Adobe is going beyond just using stock video libraries and is instead paying creators directly to submit short videos of common scenes and objects to build its training dataset. Learn More

Spotify has launched a new AI-powered feature called "AI Playlist" that allows Spotify Premium subscribers to easily create personalized playlists by simply typing in a creative prompt, with the AI then generating a playlist tailored to that prompt. The AI Playlist feature, currently in beta for users in the UK and Australia, leverages Spotify's powerful personalization technology to deliver curated music mixes based on a user's prompt, which can reference genres, moods, artists, decades, or even emojis. Learn More

Tesla CEO Elon Musk has predicted that artificial intelligence will surpass the capabilities of the smartest human within the next year. Musk made this bold claim during a recent interview, highlighting the rapid advancements being made in the field of AI and the potential for these technologies to exceed human-level intelligence in the near future. Learn More

Yann LeCun, Meta's chief AI scientist and a Turing Award winner, argues that there is no such thing as artificial general intelligence (AGI) because human intelligence is fundamentally different and far more general than the current state-of-the-art large language models (LLMs) like Meta's LLaMA, OpenAI's GPT-3, and Google's Bard. LeCun believes that to reach human-level AI, the architecture needs to be fundamentally changed from the current text-based LLMs to an "objective-driven AI" system that learns about the physical world through sensors and video data, rather than just language, in order to develop a more comprehensive "world model". Learn More

The card game developer Champions TCG has openly embraced using an AI artist to generate the artwork for their cards, claiming the AI artist can produce higher quality art than human artists and that they "don't care how he makes it, we only care that the end user enjoys our game." The company has even challenged human artists to try to match the quality of the AI-generated art within a 48-hour timeframe, offering a job as an assistant to anyone who can, indicating their strong preference for the AI-created artwork over human-made art. Learn More

Google's parent company Alphabet is reportedly considering making a bid to acquire the marketing and customer service software company HubSpot, which has a current market value of around $35 billion. However, the article notes that there is no certainty that Alphabet will actually make an offer to acquire HubSpot, as the reports indicate Alphabet is still weighing the decision.

Thanks for reading today’s Knowledge Jar.

See you next week!

Reply

or to participate.